The Unread World

Ask a salesperson what they know about a prospect and you’ll get a tidy dossier: headcount, funding, recent hires, the tools on their website. Ask what they know about the building the prospect operates from and you’ll usually get an address.

That asymmetry is strange once you notice it. For most of the economy, the building is the business. A distribution center’s value is its dock doors and its distance to the interstate. A restaurant lives or dies on its corner. A roofing contractor’s entire market is visible from above, one flat roof at a time.

The information isn’t hidden, exactly. It’s scattered. Imagery shows what’s built and what’s changed. Parcel records show who owns what and how it’s zoned. Permits show what’s about to exist. Business listings show what operates where. Population and movement show who’s nearby and how they get there. Each source answers a narrow question. None of them answers the one you actually have.

The questions businesses ask about the physical world are almost always compound. Where should we open next is really: where do the right people live, within a reasonable drive, without a competitor already serving them, on a site we could actually lease. That’s five datasets and a judgment call.

Until recently, answering a question like that meant a consultant, a GIS analyst, and a few months. Most companies didn’t bother. They expanded where the founder had a hunch, prospected from the same purchased lists as everyone else, and learned about new construction when the building opened.

What’s changed is the cost of reading. Models can now look at imagery the way an analyst would, at a scale no analyst could. Structured data that used to live in county offices is increasingly queryable. The expensive part is no longer seeing the world. It’s knowing what to look for, and what it means once you’ve found it.

That’s the work. Not the software, not the satellite. The reading.